ARTIFICIAL INTELLIGENCE Archives

22/05/2023by EOS Intelligence EOS Intelligence No Comments

Is ChatGPT Just Another Tech Innovation or A Game Changer?

447views

ChatGPT, a revolutionary AI-based conversational chatbot, has been making headlines around the world. The AI-based tool can answer user queries and generate new content in a human-like way. By automating tasks such as customer support and content creation, ChatGPT has the potential to revolutionize many industries, resulting in a more efficient digital landscape and an enhanced user experience. However, the technology is not without its risks and poses a number of issues, such as creating malicious content, copyright infringement, and other moral issues. Despite these challenges, the possibilities for ChatGPT are infinite, and with the advancement of technology, the opportunities it presents will only continue to expand.

ChatGPT is an AI-based question-and-answer chatbot that responds to user queries in a conversational way, just like how humans respond. OpenAI, a US-based research and development company, launched ChatGPT in November 2022. Since then, ChatGPT has garnered increased attention and popularity worldwide. The tool surpassed over 1 million users within five days and 100 million users within two months of launch.

ChatGPT has become popular due to its capability to answer queries in a simple and conversational manner. The tool can perform various functions, such as generating content for marketing campaigns, writing emails, blogs, and essays, debugging code, and even solving mathematics questions.

OpenAI’s ChatGPT works on the concept of generative AI and uses a language model called GPT3 – a third-generation Generative Pre-trained Transformer. The AI chatbot has been fed with about 45 terabytes of text data on a diverse range of topics from sources such as books, websites, and articles and has been trained on a set of algorithms to understand relationships between words and phrases and how it is used in context. This way, the model is able to develop an understanding of languages and generate answers. ChatGPT uses a dialog format, asks follow-up questions for clarification, admits mistakes, and is capable of dismissing inappropriate or dangerous requests.

ChatGPT also has a simple user interface, allowing communication through a plain textbox just like a messaging app, thus making it easy to use. Currently, ChatGPT is in beta testing, and users can use it for free to try and provide feedback. However, the free version is often inaccessible and out of capacity due to the increasing traffic.

In February 2023, OpenAI launched a pilot subscription plan named ChatGPT Plus, starting at US$20 per month, which is available to its customers in the USA. The subscription plan provides access to ChatGPT even during peak times and provides prior access to any new features. OpenAI is also testing ChatGPT to generate videos and pictures using its DALLE image-generating software, which is another AI tool developed by OpenAI to create art and images from text prompts. OpenAI also plans to launch a ChatGPT mobile app soon.

How could ChatGPT help businesses?

One of the most impactful areas where ChatGPT can make a difference is customer support. The AI tool can handle a large volume of consumer queries within a short time frame and give accurate responses, which can boost work efficiency and reduce employees’ workload.

In addition, the tool can also be employed to answer sales-related queries. By training ChatGPT to understand product information, pricing, and other details, businesses can provide a seamless sales experience for customers. ChatGPT can also analyze user data and behavior and can assist customers to find the products they are looking for, and give product recommendations leading to a more tailored and enjoyable shopping experience. ChatGPT can be incorporated into websites to engage visitors and help them find the information they need, which can help in lead generation.

Another potential benefit of ChatGPT is its ability to automate content generation. ChatGPT can generate unique and original content quickly, making it an effective tool for creating marketing materials such as email campaigns, blogs, newsletters, etc.

ChatGPT could be used in a number of industries, such as travel, education, real estate, healthcare, information technology, etc. For instance, in the tech industry, ChatGPT can write programs in specific programming languages such as JavaScript, Python, and React, and can be very helpful to developers in generating code snippets and for code debugging.

In healthcare, the tool can be used in scheduling appointments, summarizing patient’s health information based on previous history, assisting in diagnostics, and for telemedicine services.

In the education sector, ChatGPT can be used to prepare teaching materials and lessons and to provide personalized tutoring classes.

These are just a few applications of ChatGPT. As generative technology continues to evolve, there may be many other potential applications that can help businesses achieve their goals more efficiently and effectively.

Is ChatGPT Just Another Tech Innovation or A Game Changer by EOS Intelligence

ChatGPT’s output may not be always accurate

While ChatGPT offers several benefits and advantages, the tool is not without limitations. ChatGPT works on pre-trained data that cannot handle nuances or other ambiguities and thus may generate answers that are incorrect, biased, or inappropriate.

Moreover, ChatGPT is not connected to the internet and cannot refer to an external link to respond to queries that are not part of its training. It also does not cover the news and events after 2021 and cannot provide real-time information.

Another major limitation is that the tool is often out of capacity due to the high traffic, which makes it inaccessible. There are also other potential risks associated with these generative AI tools. Some of the threats include writing phishing emails, copyright infringement, generating abusive content or malicious software, plagiarism, and much more.

ChatGPT is not the first or only AI chatbot

While ChatGPT has garnered most of the attention in the last few months, it is neither the first nor the only AI-based chatbot in the market. There are many AI-based writers and AI chatbots in the market. These tools vary in their applications and have their own strengths and weaknesses.

For instance, ChatSonic, first released in 2020, is an AI writing assistant touted as the top ChatGPT alternative. This AI chatbot is supported by Google, has voice dictation capabilities, can generate up-to-date content, and can also generate images based on text prompts. However, ChatSonic has word limits in its free as well as paid versions, which makes it difficult for users who need to generate large pieces of text.

Similarly, Jasper is another AI tool launched in 2021, which works based on the language model (GPT-3) similar to ChatGPT. Jasper can write and generate content for blogs, videos, Twitter threads, etc., in over 50 language templates and can also check for grammar and plagiarism. Jasper AI is specifically built for dealing with business use cases and is also faster and more efficient and generates more accurate results than ChatGPT.

YouChat is another example, developed in 2022 by You.com, and running on OpenAI GPT-3. It performs similar functions as ChatGPT – responding to queries, solving math equations, coding, translating, and writing content. This chatbot cites source links of the information and acts more like an AI-powered search engine. However, YouChat lacks an aesthetic appeal and may generate results that are outdated at times.

ChatGPT-styled chatbots to power search engines

While a lot of buzz has been created about this technology, the impact of AI-based conversational chatbots is yet to be seen on a large scale. Many proclaim that tools such as ChatGPT will replace the traditional search method of using Google to obtain information.

However, experts argue that it is highly unlikely. While AI chatbots can mimic human-like conversation, they need to be trained on massive amounts of data to generate any kind of answers. These tools work on pre-trained models that were fed with large amounts of data sourced from books, articles, websites, and many more resources to generate content. Hence, real-time learning and answering would be cost-intensive in the long run.

Moreover, ChatGPT’s answers may not always be comprehensive or accurate, requiring human supervision. ChatGPT may also not be very good at solving logical questions. For instance, when asked to solve a simple problem – “RQP, ONM, _, IHG, FED, find the missing letters”, ChatGPT answered incorrectly as “LKI”. Similarly, when provided a text prompt, “The odd numbers in the group 17, 32, 3, 15, 82, 9, 1 add up to an even number”, the chatbot affirmed it, which is false. Moreover, the AI chatbot does not cover news after 2021, and when asked, “Who won the 2022 World Cup?” ChatGPT said the event has not taken place.

On the other hand, Google uses several algorithms to rank web pages and gives the most relevant web results and comprehensive information. Google has access to a much larger pool of data and the ability to analyze it in real time. Additionally, Google’s ranking algorithms have been developed over years of research and refinement, making them incredibly efficient and effective at delivering high-quality results. Therefore, while AI chatbots can be useful in certain contexts, they are unlikely to replace traditional search methods, such as Google.

However, leading search engines are looking to incorporate ChatGPT into their search tools. For instance, Microsoft is planning to incorporate ChatGPT 4, a faster version of the current ChatGPT version, into its Bing Search engine. Since 2019, the company has invested about US$13 billion in OpenAI, the parent company of ChatGPT.

In February 2023, Microsoft also incorporated ChatGPT into its popular office software Teams. With this, users with Teams premium accounts will able to generate meeting notes, access recommended tasks, and would be able to see personalized highlights of the meeting using ChatGPT. These add immense value to the user.

In February 2023, China-based e-commerce company Alibaba also announced its plan to launch its own AI chatbot similar to ChatGPT. Similarly, Baidu, a China-based internet service provider, launched a chatbot named “Ernie” in its search engine in March 2023.

Amidst the increasing popularity of ChatGPT, Google has also started working on a chatbot named “Bard” based on its own language model, Lambda. The company is planning to launch more than 20 new AI-based products in 2023. In February 2023, Google invested about US$400 million in Anthropic AI, a US-based artificial intelligence startup, which is testing a new chatbot named Claude. Thus, the race to build an effective AI-enabled search engine has just begun, and things have to unfold a bit to learn more about how chatbots can modify web searches.

On the other hand, AI technologies such as ChatGPT are sure to leave an impact on how businesses operate. With the global economy slowing down, resulting in low business margins, many businesses are looking to cut down costs to increase profitability.

ChatGPT could be extremely beneficial to companies looking to automate various business tasks, such as customer support and content generation. The tool can be integrated into channels, including websites and voice assistants. While this sounds beneficial, there is also a likelihood of the technology displacing some jobs such as customer service representatives, copywriters, research analysts, etc.

However, ChatGPT will not be replacing the human workforce completely since many business tasks require creative and critical thinking skills and other traits such as empathy and emotional intelligence that only humans have. This technology is expected to pave the way for new opportunities in various fields, such as software engineering and data analysis, and allow employees to focus on more value-added tasks instead of routine, mundane tasks, ultimately boosting productivity.

EOS Perspective

With their remarkable ability to generate human-like conversations and high-quality content, generative AI tools, such as ChatGPT, are sure to be touted as a game-changer for many businesses. The advancements in generative AI are expected to have a significant impact on various business tasks such as customer support, content creation, data analysis, marketing and sales, and even decision-making.

Investors are slowly taking note of the immense potential the technology holds. It is estimated that generative AI start-ups received equity funding totaling about US$2.6 billion across 110 deals in 2022, which echoes an increasing interest in the technology.

The adoption of generative AI technologies is poised to increase, especially in business processes where a human-like conversation is desirable. Industries such as e-commerce, retail, and travel are likely to embrace this technology to automate customer service tasks, reduce costs, and increase efficiency. In addition, generative AI is likely to become an indispensable part of industries such as finance and logistics, where high levels of accuracy and precision are required. Media and entertainment companies can also benefit from this technology to quickly generate content such as articles, videos, and audio.

That being said, generative AI is not without its risks, and the technology could be used to create fake and other discriminatory information. Hence, there is an inevitable need to ensure that generative AI models are trained and deployed in an ethical and responsible manner. Despite these challenges, there is increased research and significant activity going on in the field of generative AI, especially with regard to combining the capabilities of chatbots and traditional search engines.

The current chatbots will continue to evolve and will lead to the creation of even more advanced and sophisticated models. The popularity of generative AI tools such as ChatGPT is unlikely to wane, and the technology is here to stay, with the potential to create better prospects for business and a brighter future for society.

28/04/2023by EOS Intelligence EOS Intelligence No Comments

Automotive Industry Gearing towards Digital Transformation with AI

674views

Artificial intelligence (AI) has become an integral part of almost every industry, and the automotive sector is no exception. From self-driving cars to predictive maintenance, AI is evolving as a major disruptor in the auto industry, slowly transforming how automobiles are designed, manufactured, and sold. This digital swing is driven mainly by increased competition, consumer preferences for smart mobility, and the benefits of AI. However, AI adoption in the automotive industry is not mainstream yet, with the technology deployed only at the pilot level and in selective business segments. As the world gears toward an era of digital transformation and automation, AI is expected to be part of various business processes in the automotive industry in the coming years.

Artificial intelligence in the auto industry is typically associated with autonomous and self-driving cars. However, the technology has increasingly found its way into other applications over the last few years. Leading auto OEMs are showing an interest in deploying AI-driven innovations across the value chain, investing in tech start-ups, partnering with software providers, and building new business entities.

For instance, a venture capital fund owned by Japanese automaker Toyota, Toyota AI Ventures (rebranded as Toyota Ventures now), with US$200 million in assets under management, invested in almost 35 early-age startups that focus on AI, autonomy, mobility, and robotics between 2017 and 2020. Similarly, in 2022, South Korean automotive manufacturer Hyundai invested US$424 million to build an AI research center in the USA to advance research in AI and robotics. In the same year, CARIAD, a software division of the Germany-based Volkswagen Group, acquired Paragon Semvox GmbH, a Germany-based company that develops AI-based voice control and smart assistance systems, for US$42 million.

Changing consumer preferences, competitive pressures, and various advantages of AI are driving this transformation. According to a 2019 Capgemini research study, nearly 25% of auto manufacturers in the USA implemented AI solutions at scale, followed by the UK (14%) and Germany (12%) by the end of 2019.

There are numerous applications of AI in the automotive industry. Some of the more common and innovative uses of AI include virtual simulation models, inventory management, quality control of parts and finished goods, automated driver assistance systems (ADAS), predictive maintenance, and personalized vehicles, to name a few.

Automotive Industry Gearing towards Digital Transformation with AI by EOS Intelligence

AI-based virtual simulation models used for effective R&D processes

Due to changing customer preferences, increasing regulations concerning safety and fuel emissions, and technological disruption, OEMs are finding it more expensive to make cars nowadays. A 2020 report by PricewaterhouseCoopers says that conceptualization and product development account for 77% of the cost and 65% of the time spent in a typical automotive manufacturing process.

To make R&D cost-effective and more efficient, some auto manufacturers and tier-I suppliers are turning to AI. AI enables the simulation of digital prototypes, eliminating a lot of physical prototypes, thus reducing the costs and time for product development. One interesting concept that is emerging and catching attention in this area is the “digital twin”. The concept employs a virtual model mimicking an entire process or environment and its physical behavior. There are numerous uses of digital twins – in vehicle design and development, factory and supply chain simulations, autonomous driving simulations, etc. In vehicle design and development, digital twins make simulations easier, validate each step of the development in order to predict outcomes, improve performance, and identify possible failures before the product enters the production line.

For instance, in 2019, Continental, a Germany-based automotive parts manufacturing company, entered into a collaboration with a Germany-based start-up, Automotive Artificial Intelligence (AAI), to develop a modular virtual simulation program for its Automated Driver Assistance System (ADAS) application and also invested an undisclosed amount in the company. The virtual simulation program could generate phenomenal vehicle test data of 5,000 miles per hour compared to 6,500 miles of physical test driving per month, reducing both time and costs.

Many leading automotive companies are also looking to utilize this innovative concept in streamlining the entire manufacturing operations. For example, in early 2023, Mercedes-Benz announced that the company is partnering with Nvidia Technologies, a US-based technology company specializing in AI-based hardware and software, to build a digital twin of one of its automotive plants in Germany. Mercedes-Benz is hoping that the digital twin can help them monitor the entire plant and make quick changes in their production processes without interruptions.

General Motors, Volkswagen, and Hyundai use AI for smart manufacturing

Automation processes and industrial robots have been in automotive manufacturing for a long time. However, these systems can perform only programmed routine and repetitive tasks and cannot act on complex real-life scenarios.

The use of AI in automotive manufacturing makes these production processes smarter and more efficient. Some of the applications of AI in manufacturing include forecasting component failures, predicting demand for components and managing inventory, using collaborative robots for heavy material handling, etc.

For instance, General Motors, a US-based automotive manufacturing company, has been using AI-based design strategies since 2018 to manufacture lightweight vehicles. In 2019, the company also deployed an AI-based image classification tool in its robots to detect equipment failures on pilot-level experimentation.

Similarly, a Germany-based luxury car manufacturer, Audi, has been using AI to monitor the quality of spot welds since 2021 and is also planning to use AI in its wheel design process starting in 2023. In 2021, Audi’s parent company, Volkswagen, also invested about US$1 billion to bring technologies such as cloud-based industrial software, intelligent robotics, and AI into its factory operations. With this, the company aims to drive a 30% increase in manufacturing performance in its plants in the USA and Mexico by 2025.

In another instance, South Korean automotive manufacturer Hyundai uses AI to improve the well-being of its employees. In 2018, the company developed wearable robots for its workers, who spend most of their time in assembly lines. These robots can sense the type of work of employees, adjust their motions, and boost load support and mobility, preventing work-related musculoskeletal disorders. Thus, AI is transforming every facet of automobile manufacturing, from designing to improving the well-being of employees.

Companies provide more ADAS features amidst increasing competition

Automated Driver Assistance System (ADAS) is one of the powerful applications of AI in the automotive industry. ADAS are intelligent systems that aim to make driving safer and more efficient. ADAS primarily uses cameras and Lidar (Light Detection and Ranging) sensors to generate a high-resolution 360-degree view of the car and assists the driver or enables cars to take autonomous actions. Demand for ADAS is growing globally due to consumers’ rising preference for luxury, better safety, and comfort. It is estimated that by 2025, ADAS will become a default feature of nearly every new vehicle sold worldwide. ADAS is classified into 6 levels:

Level 0	No automation
Level 1	Driver assistance: the vehicle has at least a single automation system
Level 2	Partial driving automation: the vehicle has more than one automated system; the driver has to be on alert at all times
Level 3	Conditional driving automation: the vehicle has multiple driver assistance functions that control most driving tasks; the driver has to be present to take over if anything goes wrong
Level 4	High driving automation: the vehicle can make decisions itself in most circumstances; the driver has the option to manually control the car
Level 5	Full driving automation: the vehicle can do everything on its own without the presence of a driver

At present, cars from level 0 to level 2 are on the market. To meet the growing competitive edge, several auto manufacturers are adding more automation features to the level 2 type. Companies have also been making significant strides toward developing autonomous vehicles. For instance, auto manufacturers such as Mercedes, BMW, and Hyundai are testing level 3 autonomous vehicles, and Toyota and Honda are testing and trialing level 4 vehicles. This indicates that the future of mobility will be highly automated relying upon technologies such as AI.

Volkswagen and Porsche use AI in automotive marketing and sales

There are various applications of AI in marketing and sales operations – in sales forecasting and planning, personalized marketing, AI-assisted virtual assistants, etc. According to a May 2022 Boston Consulting Group (BCG) report, auto OEMs can gain faster returns with lower investments by deploying AI in their marketing and sales operations.

Some automotive companies have already started to deploy AI in sales and marketing. For instance, since 2019, Volkswagen has been leveraging AI to create precise market forecasts based on certain variables and uses the data for its sales planning. Similarly, in 2021, a Germany-based luxury car manufacturer, Porsche, launched an AI tool that suggests various vehicle options and their prices based on the customer’s preferences.

Automakers integrate AI-assisted voice assistants into cars

Cars nowadays are not only perceived as a means of transportation, but consumers also expect sophisticated features, convenience, comfort, and an enriching experience during their journey. AI enhances every aspect of the cockpit and deploys personalized infotainment systems that learn from user preferences and habits over time. Many automakers are integrating AI-based voice assistants to help drivers navigate through traffic, change the temperature, make calls, play their favorite music, and more.

For instance, in 2018, Mercedes-Benz introduced the Mercedes Benz User Experience (MBUX) voice-assisted infotainment system, which gets activated with the keyword “Hey Mercedes”. Amazon, Apple, and Google are also planning to get carmakers to integrate their technologies into in-car infotainment systems. It is expected that 90% of new vehicles sold globally will have voice assistants by 2028.

Integration and technological challenges hamper the adoption of AI

The adoption of AI in the automotive industry is still at a nascent stage. Several OEM manufacturers in the automotive industry are leveraging various AI solutions only at the pilot level, and scaling up is slow due to the various challenges associated with AI.

At the technology level, the creation of AI algorithms remains the main challenge, requiring extensive training of neural networks that rely on large data sets. Organizations lack the skills and expertise in AI-related tools to successfully build and test AI models, which is time-consuming and expensive. AI technology also uses a variety of high-priced advanced sensors and microprocessors, thus hindering the technology from being economically feasible.

Moreover, AI acts more or less like a black box, and it remains difficult to determine how AI models make decisions. This obscurity remains a big problem, especially for autonomous vehicles.

At the organizational level, integration challenges make it difficult to implement the technology with existing infrastructure, tools, and systems. Lack of knowledge of selecting and investing in the right AI application and lack of information on potential economic returns are other biggest organizational hurdles.

EOS Perspective

The applications of AI in the automotive industry are broad, and many are yet to be envisioned. There has been an upswing in the number of automotive AI patents since 2015, with an average of 3,700 patents granted every year. It is evident that many disrupting high-value automotive applications of AI are likely to be deployed in the coming decade. Automotive organizations are bolstering their AI skills and capabilities by investing in AI-led start-ups. These companies together already invested about US$11.2 billion in these startups from 2014 to 2019.

There is also an increase in the hiring pattern of AI-related roles in the industry. Many automotive industry leaders are optimistic that AI technology can bring significant economic and operational benefits to their businesses. AI can turn out to be a powerful steering wheel to drive growth in the industry. The future of many industries will be digital, and so will be for the automotive sector. Hence, for automotive businesses that are yet to make strides toward this digital transformation, it is better to get into this trend before it gets too late to keep up with the competition.

07/09/2022by EOS Intelligence EOS Intelligence No Comments

Powering Healthcare Diagnostics with AI: a Pipe Dream or Reality

692views

The growing paucity of radiologists across the globe is alarming. The availability of radiologists is extremely disproportionate globally. To illustrate this, Massachusetts General Hospital in Boston, USA, had 126 radiologists, while the entire country of Liberia had two radiologists, and 14 countries in the African continent did not have a single radiologist, as of 2015. This leads to a crucial question – how to address this global unmet demand for radiologists and diagnostic professionals?

Increasing capital investment signals rising interest in AI in healthcare diagnostics

The global market for Artificial Intelligence (AI) in healthcare diagnostics is forecast to grow at a CAGR of 8.3%, from US$513.3 million in 2019 to US$825.9 million in 2025, according to Frost & Sullivan’s report from 2021. This growth in the healthcare diagnostics AI market is attributed to the increased demand for diagnostic tests due to the rising prevalence of novel diseases and fast-track approvals from regulatory authorities to use AI-powered technologies for preliminary diagnosis.

Imaging Diagnostics, also known as Medical Imaging is one of the key areas of healthcare diagnostics that is most interesting in exploring AI implementation. From 2013 to 2018, over 70 firms in the imaging diagnostics AI sector secured equity funding spanning 119 investment deals and have progressed towards commercial beginnings, thanks to quick approvals from respective regulatory bodies.

Between 2015 and 2021, US$3.5 billion was secured by AI-enabled imaging diagnostics firms (specialized in developing AI-powered solutions) globally for 290 investment deals, as per Signify Research. More than 200 firms (specialized in developing AI-powered solutions) globally were building AI-based solutions for imaging diagnostics, between 2015 and 2021.

The value of global investments in imaging diagnostics AI in 2020 was approximately 8.8% of the global investments in healthcare AI. The corresponding figure in 2019 was 10.2%. The sector is seeing considerable investment at a global level, with Asia-based firms (specialized in developing AI-powered solutions) having secured around US$1.5 billion, Americas-based companies raising US$1.2 billion, and EMEA-based firms securing over US$600 million between 2015 and 2021.

As per a survey conducted by the American College of Radiology in 2020 involving 1,427 US-based radiologists, 30% of respondents said that they used AI in some form in their clinical practice. This might seem like a meager adoption rate of AI amongst US radiologists. However, considering that five years earlier, there were hardly any radiologists in the USA using AI in their clinical practice, the figure illustrates a considerable surge in AI adoption here.

However, the adoption of AI in healthcare diagnostics is faced with several challenges such as high implementation costs, lack of high-quality diagnostic data, data privacy issues, patient safety, cybersecurity concerns, fear of job replacement, and trust issues. The question that remains is whether these challenges are considerable enough to hinder the widespread implementation of AI in healthcare diagnostics.

AI advantages help answer the needs in healthcare diagnostics

Several advantages such as improved correctness in disease detection and diagnosis, reduced scope of medical and diagnosis errors, improved access to diagnosis in areas where radiologists are unavailable, and increased workflow and efficacy drive the surge in the demand for AI-powered solutions in healthcare diagnostics.

One of the biggest benefits of AI in healthcare diagnostics is improved correctness in disease detection and diagnosis. According to a 2017 study conducted by two radiologists from the Thomas Jefferson University Hospital, AI could detect lesions caused by tuberculosis in chest X-rays with an accuracy rate of 96%. Beth Israel Deaconess Medical Center in Boston, Massachusetts uses AI to scan images and detect blood diseases with a 95% accuracy rate. There are numerous similar pieces of evidence supporting the AI’s ability to offer improved levels of correctness in disease detection and diagnosis.

A major benefit offered by AI in healthcare diagnostics is the reduced scope of medical and diagnosis errors. Medical and diagnosis errors are among the top 10 causes of death globally, according to WHO. Taking this into consideration, minimizing medical errors with the help of AI is one of the most promising benefits of diagnostics AI. AI is capable of cutting medical and diagnosis errors by 30% to 40% (trimming down the treatment costs by 50%), according to Frost & Sullivan’s report from 2016. With the implementation of AI, diagnostic errors can be reduced by 50% in the next five years starting from 2021, according to Suchi Saria, Founder and CEO, Bayesian Health and Director, Machine Learning and Healthcare Lab, Johns Hopkins University.

Another benefit that has been noticed is improved access to diagnosis in areas where there is a shortage of radiologists and other diagnostic professionals. The paucity of radiologists is a global trend. To cite a few examples, there is one radiologist for: 31,707 people in Mexico (2017), 14,634 people in Japan (2012), 130,000 people in India (2014), 6,827 people in the USA (2021), 15,665 people in the UK (2020).

AI has the ability to modify the way radiologists operate. It could change their active approach toward diagnosis to a proactive approach. To elucidate this, instead of just examining the particular condition for which the patient requested medical intervention, AI is likely to enable radiologists to find other conditions that remain undiagnosed or even conditions the patient is unaware of. In a post-COVID-19 era, AI is likely to reduce the backlogs in low-emergency situations. Thus, the technology can help bridge the gap created due to radiologist shortage and improve the access to diagnosis of patients to a drastic extent.

Further, AI helps in improving the workflow and efficacy of healthcare diagnostic processes. On average at any point in time, more than 300,000 medical images are waiting to be read by a radiologist in the UK for more than 30 days. The use of AI will enable radiologists to focus on identifying dangerous conditions rather than spend more time verifying non-disease conditions. Thus, the use of AI will help minimize such delays in anomaly detection in medical images and improve workflow and efficacy levels. To illustrate this, an AI algorithm named CheXNeXt, developed in a Stanford University study in 2018 could read chest X-rays for 14 distinct pathologies. Not only could the algorithm achieve the same level of precision as the radiologists, but it could also read the images in less than two minutes while the radiologists could read them in an average of four hours.

Black-box AI: A source of challenges to AI implementation in healthcare diagnostics

The black-box nature of AI means that with most AI-powered tools, only the input and output are visible but the innards between them are not visible or knowable. The root cause of many challenges for AI implementation in healthcare diagnostics is AI’s innate character of the black box.

One of the primary impediments is tracking and evaluating the decision-making process of the AI system in case of a negative result or outcome of AI algorithms. That is to say, it is not possible to detect the fundamental cause of the negative outcome within the AI system because of the black-box nature of AI. Therefore, it becomes difficult to avoid such occurrences of negative outcomes in the future.

The second encumbrance caused by the black-box nature of AI is the trust issues of clinicians that are hesitant to use AI applications because they do not completely comprehend the technology. Patients are also expected to not have faith in the AI tools because they are less forgiving of machine errors as opposed to human errors.

Further, several financial, technological, and psychological challenges while implementing AI in healthcare diagnostics are also associated with the black-box nature of the technology.

Financial challenges

High implementation costs

According to a 2020 survey conducted by Definitive Healthcare, a leading player in healthcare commercial intelligence, cost continues to be the most prominent encumbrance in AI implementation in diagnostics. Approximately 55% of the respondents who do not use AI pointed out that cost is the biggest challenge in AI implementation.

The cost of a bespoke AI system can be between US$20,000 to US$1 million, as per Analytics Insights, while the cost of the minimum viable product (a product with sufficient features to lure early adopters and verify a product idea ahead of time in the product development cycle) can be between US$8,000 and US$15,000. Other factors that also decide the total cost of AI are the costs of hiring and training skilled labor. The cost of data scientists and engineers ranges from US$550 to US$1,100 per day depending on their skills and experience levels, while the cost of a software engineer (to develop applications, dashboards, etc.) ranges between US$600 and US$1,500 per day.

It can be gauged from these figures that the total cost of AI implementation is high enough for the stakeholders to ponder upon the decision of whether to adopt the technology, especially if they are not fully aware of the benefits it might bring and if they are working with ongoing budget constraints, not infrequent in healthcare institutions.

Technological challenges

Overall paucity of availability of high-quality diagnostic data

High-quality diagnostic and medical datasets are a prerequisite for the testing of AI models. Because of the highly disintegrated nature of medical and diagnostic data, it becomes extremely difficult for data scientists to procure the data for testing AI algorithms. To put it in simple terms, patient records and diagnostic images are fragmented across myriad electronic health records (EHRs) and software platforms which makes it hard for the AI developer to use the data.

Data privacy concerns

AI developers must be open about the quality of the data used and any limitations of the software being employed, without risking cybersecurity and without breaching intellectual property concerns. Large-scale implementation of AI will lead to higher vulnerability of the existing cloud or on-premise infrastructure to both physical and cyber attacks leading to security breaches of critical healthcare diagnostic information. Targets in this space such as diagnostic tools and medical devices can be compromised by malware or software viruses. Compromised data and algorithms will result in errors in diagnosis and consequently inaccurate recommendations of treatment thereby causing stakeholders to refrain from using AI in healthcare diagnostics.

Patient safety

One of the foremost challenges for AI in healthcare diagnostics is patient safety. To achieve better patient safety, developers of AI algorithms must ensure the credibility, rationality, and transparency of the underlying datasets. Patient safety depends on the performance of AI which in turn depends on the quality of the training data. The better the quality of the data, the better will be the performance of the AI algorithms resulting in higher patient safety.

Mental and psychological challenges

Fear of job substitution

A survey published in March 2021 by European Radiology, the official journal of the European Society of Radiology, involving 1,041 respondents (83% of them were based in European countries) found that 38% of residents and radiologists are worried about their jobs being cut by AI. However, 48% of the respondents were more enterprising and unbiased towards AI. The fear of substitution could be attributed to the fact that those having restricted knowledge of AI are not completely educated about its shortcomings and consider their skillset to be less up-to-date than the technology. Because of this lack of awareness, they fail to realize that radiologists are instrumental in developing, testing, and implementing AI into clinical practice.

Trust issues

Trusting AI systems is crucial for the profitable implementation of AI into diagnostic practice. It is of foremost importance that the patient is made aware of the data processing and open dialogues must be encouraged to foster trust. Openness or transparency that forges confidence and reliability among patients and clinicians is instrumental in the success of AI in clinical practice.

EOS Perspective

With trust in AI amongst clinicians and patients, its adoption in healthcare diagnostics can be achieved at a more rapid pace. Lack of it breeds fear of job replacement by the technology amongst clinicians. Further, scarcity of awareness of AI’s true potential as well as its limitations also threatens diagnostic professionals from getting replaced by the technology. Therefore, to fully understand the capabilities of AI in healthcare diagnostics, clinicians and patients must learn about and trust the technology.

With the multitude and variety of challenges for AI implementation in healthcare diagnostics, its importance in technology becomes all the more critical. The benefits of AI are likely to accelerate the pace of adoption and thereby realize the true potential of AI in terms of saving clinicians’ time by streamlining how they operate, improving diagnosis, minimizing errors, maximizing efficacy, reducing redundancies, and delivering reliable diagnostic results. To power healthcare diagnostics with AI, it is important to view AI as an opportunity rather than a threat. This in turn will set AI in diagnostics on its path from pipe dream to reality.

26/08/2021by EOS Intelligence EOS Intelligence No Comments

Beauty Tech Giving Beauty Industry a Facelift

1.3kviews

In recent years, artificial intelligence and virtual reality have been adding an additional dimension to the beauty industry, quite literally. With consumers increasingly embracing and demanding personalized offerings and precise results, leading brands, such as L’Oréal and Shiseido are investing heavily in the space. Just as in many other industries, AI is revolutionizing beauty products and how they are conceptualized, created, and sold. However, it is a long road from being perceived as gimmicky promotions to improving customer engagement to becoming commercial go-to solutions.

Artificial intelligence (AI) has been greatly integrated in our lives through different sectors and now the beauty industry is no exception. The use of AI, augmented reality (AR), virtual reality (VR) as well as complex beauty devices has revolutionized the way consumers perceive, apply, and select beauty products. Moreover, in the age of online retail, it enables companies to maintain a similar personalized level of service that would otherwise require a physical interaction with a beauty consultant. Technology is creating new experiences for the consumer, both in terms of beauty products’ features as well as purchasing process.

Beauty industry is also one of the most competitive sectors, with consumers always being on the lookout for new products and having low brand loyalty. Beauty tech seems to address this issue as well, as it elevates consumer engagement through enhanced personalized offerings, which in turn is a trend that has been driving the beauty industry for several years now.

The three main aspects of beauty tech encompass personalization through AI, virtual makeup using AR and VR, and smart skincare tools/beauty gadgets.

Personalization through AI

Across the retail sector, the key to consumer’s heart and pockets for a long time has been personalization of products and sales experience. Beauty industry is no exception. Consumers have been looking for the perfect skincare product that work best for them or the lipstick shade that goes perfectly with their skin tone. Moreover, consumers want this all from the comfort of their home. This is where AI comes in.

Through retail kiosks and mobile apps, AI enables companies to offer personalized shade offerings that are especially curated for the individual user. A number of companies is investing and capitalizing on this technology to differentiate themselves in the eyes of the consumer. One of the leading market players in the beauty industry, L’Oréal, has been one of the first companies to invest in AI- and VR-based beauty tech and acquired Toronto-based, ModiFace, in 2018. There are several different ways companies, such as L’Oréal, have incorporated AI into their product offerings.

Beauty Tech Giving Beauty Industry a Facelift by EOS Intelligence

Lancôme (a subsidiary of L’Oréal) has placed an AI-powered machine, called Le Teint Particulier, at Harrods and Selfridges in the UK, which creates custom-made foundation for the customer. The machine first identifies ones facial color using a handheld scanner, post which it uses a proprietary algorithm to select a foundation shade from 20,000 combinations. Following this, the machine creates the personalized shade for the user, which can then be bottled and purchased.

In addition to physical store solutions, AI-powered apps and websites also offer consumers personalized recommendations. In 2019, L’Oréal applied ModiFace’s AI technology to introduce a new digital skin diagnostic tool, called SkinConsult, for its brand, Vichy. The AI-powered tool uses more than 6,000 clinical images in order to deliver accurate skin assessment for all skin types. It analyzes selfies uploaded by users to identify fine lines, dark spots, wrinkles, and other issues, and then provides tailored product and routine recommendations to the user to address the skin concerns.

My Beauty Matches, a UK-based company, offers AI-based personalized and impartial beauty product recommendations and price comparisons. The website asks consumers diagnostic-style questions about their skin and hair type, concerns, and preferences, and uses AI to analyze the data and recommend products from 400,000 products (from about 3,500 brands) listed on its website. Alongside, the company runs Beauty Matches Engine (BME), which is a solution for beauty retailers using consumer data and AI algorithms to identify consumer purchasing and browsing patterns as well as their preferred products by age and skin or hair concerns. This helps retailers predict and stock, which product the consumer is likely to purchase, improving sales, increasing upsells, and providing a personalized solution to customers.

On similar lines, another app, Reflexion, uses AI to measure the shininess of skin through pictures and offers personalized product recommendations. The app claims to provide much deeper analysis than regular image analysis apps and provides additional features such as testing if products such as foundation are evenly applied. The app works by measuring a face surface’s Bidirectional Scatter Distribution Function (BSDF), which is a measure of light reflected on the user’s face.

Nudemeter is another such product, which uses AI to personalize makeup choices and foundation shades for a full spectrum of skin tones, including darker skins. The app uses color analysis and digital image processing along with its AI algorithms that ensure accurate color measurement irrespective of background lighting, pixels, etc. The app is currently being used by Spktrm Beauty, a US-based niche beauty company targeting shoppers with dark skin.

Virtual makeup through AR and VR

In today’s world where consumers prefer to shop from the comfort of their homes, AR and VR are enabling beauty companies to provide experience similar to that of physical retail to their consumers. AR and VR technologies-based apps let users experiment virtually with a range of cosmetics by allowing them to try several different shades, all within minutes and through their smartphone. This elevates the users shopping experience and improves sales conversion.

Sephora’s Virtual Makeup Artist enables customers to try on thousands of shades of lipsticks and eyeshadows through their smartphones or at kiosks at Sephora stores. While many such apps and filters have been in use for some time now, they are increasingly becoming more sophisticated, providing accurate color match to the skin and ensuring the virtual makeup does not move when the user shakes their face, changes to a side angle, etc. In addition, such apps also provide digital makeup tutorials to engage customers.

On similar lines, L’Oréal uses ModiFace’s AR and AI technology to provide virtual makeup try-on on Amazon and Facebook. The technology enables customers using these two platforms to try on different shades of lipsticks and other make-up products through a live video or a selfie from an array of L’Oréal brands such as Maybelline, L’Oréal Paris, NYX Professional Makeup, Lancôme, Giorgio Armani, Yves Saint Laurent, Urban Decay, and Shu Uemura.

Moreover, AR-based try-on apps helped brands connect with their customers during the previous year when most customers were stuck home and could not physically try on make-up. LVMH-owned Benefit Cosmetics has been investing in AR tech, and launched Benefit’s Brow Try-On Experience program (along with Taiwanese beauty-tech company, Perfect Corporation), which helps online shoppers identify the right eyebrow shape and style for them and then choose products accordingly. The company uses facial point detector technology for the program. The app witnessed a 43% surge in its daily users during April and May of 2020 (as compared with January and March 2020), when people were confined to their homes owing to the COVID outbreak. This helped connect with consumers in a fresh manner and increased brand loyalty. Moreover, Benefit claims that brows products have been their strongest category post-COVID outbreak.

One of China’s leading e-commerce players, Alibaba, also partnered with Perfect Corporation to integrate the latter’s ‘YouCam Makeup’ (an AR-based virtual makeup try-on technology) into Alibaba’s Taobao and Tmall online shopping experience.

Smart devices

In addition to AI and AR based apps and solutions, smart devices is another category in the beauty tech space that is gaining momentum. A certain section of premium consumers are increasingly open to invest heavily into smart beauty gadgets that not only improve skin and hair quality but also help them quantitatively measure the results from using a certain product. While these products are currently expensive and for a niche audience, they have been gaining popularity, especially across the USA and China.

One such smart skincare device is L’Oréal’s Perso, which is based on ModiFace’s AI-powered skin diagnostics and analysis technology. Perso uses AI, location data, and consumer preferences to formulate personalized moisturizer for the consumer. The product is further expected to extend into foundations and lip shades. Perso is expected to be launched in 2021.

On similar lines, in July 2019, Japan-based Shiseido, launched its smart skincare device called Optune, which measures a user’s location-based weather and air pollution data, sleep data, stress levels, and menstrual cycles to create a custom moisturizer. Optune is available on a subscription basis and costs about US$92 per month.

In 2020, P&G also launched a premium skincare system, called Opte Precision. The skincare device uses blue LED light to scan one’s skin and applies a patented precision algorithm to detect problem areas and analyze complexion. Post this, the device releases an optimizing serum that is applied to spots to instantly cover age spots, pigmentation, etc., and to fade their appearance over time. The device has 120 nozzles and works on a technology similar to that of a thermal inkjet printer. The device targets a premium niche audience and costs US$599 with refill cartridge costing US$100.

In 2018, Johnson & Johnson’s drugstore skincare brand, Neutrogena, also launched a smart skincare device – a skin scanner, called Skin360 and SkinScanner, which uses technology from FitSkin (a US-based technology company). The scanner comes in the form of a magnifying camera that gets attached to a smartphone. The camera, which has a 30-time magnifying power helps scan the size and appearance of one’s pores, size and depth of fine lines and wrinkles, the skin’s moisture level, and also provides a score to the skin’s hydration level. The data is processed in a mobile app, which in turn provides a complete skin analysis and offers expert advice and product recommendations. While most smart skin devices are relatively expensive, this one retails at around US$50.

EOS Perspective

While AI and AR have been embraced by a lot of industries in the past, beauty tech is still in its infancy. That being said, there is a lot of potential in the space, especially with the consumer becoming increasingly comfortable with technology. While till recently, most technology-based products in the beauty sector were gimmicky and more for fun and consumer engagement, brands have started taking this space seriously, and started launching products that offer real sales growth opportunity.

Moreover, while AI and AR-based technologies have been accepted fairly easily by the consumers and industry players alike, smart devices is still a very niche category, with most products focused on a niche affluent clientele, who are willing to spend more than US$100 on products that may help improve their skin. There is a lot of potential for this segment to innovate, collaborate, and launch products at a more affordable price point in order to reach the masses.

Over the next couple of years, we can expect new niche players, exploring the benefits of beauty tech to enter the market in addition to greater number of partnerships between traditional beauty giants and technology companies. As personalization continues to be the mantra for consumers, beauty companies cannot look to ignore the space in the coming future.

14/10/2020by EOS Intelligence EOS Intelligence No Comments

Agritech in Africa: How Blockchain Can Help Revolutionize Agriculture

4.5kviews

In the first part of our series on agritech in Africa, we took a look into how IT and other technology investments are helping small farmers in Africa. In the second part, we are exploring the impact that potential application of advanced technologies such as blockchain can have on the African agriculture sector.

Blockchain, or distributed ledger technology, is already finding utility across several business sectors including financial, banking, retail, automotive, and aviation industries (click here to read our previous Perspectives on blockchain technology). The technology is finding its way in agriculture too, and has the potential to revolutionize the way farming is done.

This article is the second part of a two-piece coverage focusing on technological advancements in agriculture across the African continent.

Read part one here: Agritech in Africa: Cultivating Opportunities for ICT in Agriculture

State of blockchain implementation in agriculture in Africa

Agricultural sector in Africa has already witnessed the onset of blockchain based solutions being introduced in the market. Existing tech players and emerging start-ups have developed blockchain solutions, such as eMarketplaces, agricultural credit/financing platforms, and crop insurance services. Companies, globally as well as within Africa, are harnessing applications of blockchain to develop innovative solutions targeted at key stakeholders across the food value chain.

Blockchain to promote transparency across agriculture sector

The most common application of blockchain in any industry sector (and not only agriculture) is creating an immutable record of transactions or events, which is particularly helpful in creating a trusted record of land ownership for farmers, who are traditionally dependent on senior village officials to prove their ownership of land.

Since 2017, a Kenyan start-up, Land LayBy has been using an Ethereum-based shared ledger to keep records of land transactions. This offers farmers a trusted and transparent medium to establish land ownership, which can then further be used to obtain credit from banks or alternative financing companies. BanQu and BitLand are other examples of blockchain being used as a proof of land ownership.

This feature of blockchain also enables creation of a transparent environment where companies can trace the production and journey of agricultural products across their supply chain. Transparency across the supply chain helps create trust between farmers and buyers, and the improved visibility of prices further down the value chain also enables farmers to get better value for their produce.

In 2017, US-based Bext360 started a pilot project with US-based Coda Coffee and its Uganda-based coffee export partner, Great Lakes Coffee. The company developed a machine to grade and weigh coffee beans deposited to Great Lakes by individual farmers in East Uganda. The device uploads the data on a blockchain-based SaaS solution, which enables users to trace the coffee from its origin to end consumer. The blockchain solution is also used to make payments to the farmers based on the grade of their produce in form of tokens.

In 2017, Amsterdam-based Moyee Coffee also partnered with KrypC, a global blockchain, to create a fully blockchain-traceable coffee. The coffee beans are sourced from individual farmers in Ethiopia, and then roasted within the country, before being exported to the Netherlands.

This transparency can help food companies to isolate the cause of any disease outbreak impacting the food value chain. This also allows consumers can be aware of the source of the ingredients used in their food products.

Blockchain-based platforms to improve farmer and buyer collaboration

Blockchain can also act as a platform to connect farmers with vendors, food processing, and packaging companies, providing a secure and trusted environment to both buyers and suppliers to transact without the need of a middleman. This also results in elimination of margins that need to be paid to these intermediaries, and helps improve the margins for buyers.

Farmshine, a Kenyan start-up, created a blockchain-based platform to auger trade collaboration among farmers, buyers, and service providers in Kenya. In January 2020, the company also raised USD$250,000 from Gray Matters Capital, to finance its planned future expansion to Malawi.

These blockchain platforms can also be used to connect farmers to other farmers, for activities such as asset or land sharing, resulting in more efficiency in economical farming operations. Blockchain platform can also enable small farmers to lease idle farms from their peers, thereby providing them with access to additional revenue sources, which they would not be able to do traditionally.

AgUnity, an Australian-start-up established in 2016, developed a mobile application which enables farmers to record their produce and transactions over a distributed ledger, offering a trusted and transparent platform to work with co-operatives and third-party buyers. The platform also enables farmers to share farming equipment as per a set schedule to improve overall operational and cost efficiency. In Africa, AgUnity has launched pilot projects in Kenya and Ethiopia, targeted at helping farmers achieve better income for their produce.

A Nigerian start-up, Hello Tractor uses IBM’s blockchain technology to help small farmers in Nigeria, which cannot afford tractors on their own, to lease idle tractors from owners and contractors at affordable prices through a mobile application.

Smart contracts to transform agriculture finance and insurance

Less than 3% of small farmers in sub-Saharan Africa have adequate access to agricultural insurance coverage, which leaves them vulnerable to adverse climatic situations such as droughts.

Smart contracts based on blockchain can also be used to provide crop-insurance, which can be triggered given certain set conditions are met, enabling farmers to secure their farms and family livelihood in case of extreme climatic events such as floods or droughts.

SmartCrop, an Android-based mobile platform, provides affordable crop insurance to more than 20,000 small farms in Ghana, Kenya, and Uganda through blockchain-based smart contracts, which are triggered based on intelligent weather predictions.

Netherlands-based ICS, parent company of Agrics East Africa (which provides farm inputs on credit to small farmers in Kenya and Tanzania) is also exploring a blockchain-wallet based saving product, “drought coins”, which can be encashed by farmers depending on the weather conditions and forecasts.

Tracking of assets (such as land registries) and transactions on the blockchain can also be used to verify the farmers’ history, which can be used by alternative financing companies to offer loans or credits to farmers – e.g. in cases when farmers are not able to get such financing from traditional banks – transforming the banking and financial services available to farmers.

Several African start-ups such as Twiga Foods and Cellulant have tried to explore the use of blockchain technology to offer agriculture financing solutions to small farmers in Africa.

In late 2018, Africa’s leading mobile wallet company, Cellulant, launched Agrikore, a blockchain-based digital-payment, contracting, and marketplace system that connects small farmers with large commercial customers. The company started its operations in Nigeria and is exploring expansion of its business to Kenya.

In 2018, Kenya-based Twiga Foods (that connects farmers to urban retailers in an informal market) partnered with IBM to launch a blockchain-based lending platform which offered loans to small retailers in Kenya to purchase food products from suppliers listed on Twiga platform.

Read our previous Perspective Africa’s Fintech Market Striding into New Product Segments to find out more about innovative fintech products for agriculture and other sectors financing in Africa

And last, but not the least, blockchain or cryptocurrencies can simply be used as a mode of payment with a much lower transaction fee offered by traditional banking institutions.

Improving mobile internet access to boost blockchain implementation

While blockchain has shown potential to transform agriculture in Africa, its implementation is limited by the lack of mobile/internet access and technical know-how among small farmers. As of 2018, mobile internet had penetrated only 23% of the total population in Sub-Saharan Africa.

However, the GSM Association predicts mobile internet penetration to improve significantly over the next five years, to ~39% by 2025. Improved access to internet services is expected to boost the farmers’ ability to interact with the blockchain solutions, thereby increasing development and deployment of more blockchain-based solutions for farmers.

EOS Perspective

Agritech offers an immense opportunity in Africa, and blockchain is likely to be an integral part of this opportunity. Blockchain has already started witnessing implementation in systems providing proof of ownership, platforms for farmer cooperation, and agricultural financing tools.

Unlike Asian and Latin American countries, African markets have shown a relatively positive attitude towards adoption of blockchain, a fact that promises positive environment for development of such solutions.

At the moment, most development in blockchain agritech space is concentrated in Kenya, Nigeria, Uganda, and Ghana. However, there is potential to scale up operations in other countries across Africa as well, and some start-ups have already proved this (e.g. Farmshine was able to secure the necessary financing to expand its presence in Malawi). Other companies can follow suit, however, that would only be possible with the help of further private sector investments.

Still in the nascent stages of development, blockchain solutions face an uncertain future, at least in the short term, and are dependent on external influences to pick up growth they need to impact the agriculture sector significantly. However, once such solutions achieve certain scalability, and become increasingly integrated with other technologies, such as Internet of Things and artificial intelligence, blockchain has the capability of completely transform the way farming is done in Africa.

17/09/2020by EOS Intelligence EOS Intelligence No Comments

Agritech in Africa: Cultivating Opportunities for ICT in Agriculture

2.7kviews

Agriculture technologies in Africa have been undergoing significant development over the years, with many tech start-ups innovating information and communications technologies to support agriculture at all levels. While some technologies have been successfully launched, some are in initial stages of becoming a success. Private sector investments have been the key driving factor supporting the development of agriculture technologies in Africa. In the first part of our series on agritech in Africa, we are examine what impact and opportunities arise from the use of these technologies in Africa.

Agriculture plays a significant role in Africa’s economy, contributing 32% to the continent’s GDP and employing 65% of the total work force (as per the World Bank estimates). Nearly 70% of the continent’s population directly depends on agribusiness. Vast majority of farmers work on small scale farms that produce nearly 90% of all agricultural output.

This article is the first part of a two-piece coverage focusing on technological advancements in agriculture across the African continent.

Read part two here: Agritech in Africa: How Blockchain Can Help Revolutionize Agriculture

Agriculture in Africa has been under the pressure of many challenges such as low productivity, lack of knowledge and exposure to new farming techniques, and lack of access to financial support, especially for the small-scale farmers. These challenges are prompting investments in newer technologies to enhance the productivity through smart agriculture techniques.

Lately, there have been an increased use of various technologies in agriculture in Africa, such as Internet of Things (IoT), Open Source Software, Cloud Computing, Artificial Intelligence, Drones/Unmanned Aerial Vehicles (UAVs), and Big Data Analytics. Many tech start-ups have developed solutions targeting various aspects of agriculture, including finance, supply chain, retailing, and even delivering information related to crops and weeds. These solutions are accessible to farmers through front-end devices such as smart phones and tablets, or even SMS.

Start-ups lead agritech development in Africa

Many agritech start-ups in Africa have come up with solutions that have led to a rise in productivity of the farms. Drones have been a breakthrough technology, helping farmers oversee their crops, and manage their farms effectively. Drones use highly focused cameras to capture picture of crops, soil or weeds. This, coupled with big data analytics and Artificial Intelligence (AI), provides insights to farmers, saving their time and effort, while also helping them find potential issues which could impact the productivity of their farms.

There are various agritech start-ups that are developing such drones, and providing them to farmers for rent or lease to analyse their crops and farms. A South African agritech start-up, Aerobotics, offers an end-to-end solution to help farmers manage their farms using drones, through early detection of any crop-related problems, and offering curative measures for the problems using an AI-based analytics platform. The company partners with drone manufacturing companies such as DJI and Micasense to deliver these solutions.

Acquahmeyer, another start-up based in Ghana, also provides drones to its farming customers to help them use a comprehensive approach to apply crop pest control and plant nutrition management for their farms.

Advent of advanced technologies such as IoT is also helping farmers to adopt smart farm management through the use of smart sensors connected in a network. This helps every farmer to get granular details of the crops, soil, farming equipment, or livestock, enabling the farmers to devise appropriate farming approaches.

Kenya-based UjuziKilimo provides solution for analyzing soil characteristics using electronic sensor placed in the ground. This helps farmers with useful real-time insights into soil conditions. The solution further utilizes big data analytics to guide the farmers, by offering insights through SMS on their connected mobile phones or tablets.

Hello Tractor, a Kenyan start-up, provides an IoT solution, through which farmers can have access to affordable tractors which are monitored virtually through a remote asset tracking device on the tractor, sharing data over the Hello Tractor Cloud. Farmers, booking agents, dealers, and tractor owners are connected via IoT. The company is also collaborating with IBM to incorporate artificial intelligence and blockchain to their solutions.

AI has also witnessed a rapid growth in adoption across agriculture sector in Africa. Agrix Tech, based in Cameroon, has developed a mobile application that requires the farmers to capture the picture of diseased crop, which is then analyzed via AI to detect crop diseases, and helps the farmers with treatment solution to save their crops.

AI is also helping Kenyan farmers with the knowledge on planting the right crops at the right time. Tech giant, Capgemini, has teamed up with a Kenyan social enterprise in Kakamega region in Western Kenya to use artificial intelligence to analyze farming data, and then send insights about right time and technique of planting crops to the farmers’ cell phones.

There are other agritech solutions that include mobile applications which use digital platforms such as cloud computing to reach out to farmers, and provide them with apt agriculture solutions. Ghana-based CowTribe offers a mobile USSD-based subscription service which enables livestock farmers to connect with veterinarians for animal vaccines and other livestock healthcare services using cloud-based logistics management system. The company focuses on managing the schedules, and delivering the right service to the livestock farmers, to help them safeguard their animals from any health-related problems.

Several agritech investments are also impacting the financial side of agriculture. Kenya-based Apollo Agriculture provides solutions related to financing, farm inputs, advice insurance and market access through the use of agronomic machine learning, remote sensing, and mobile technology using satellite data and cloud computing.

Another Nigerian start-up Farmcrowdy has developed Nigeria’s first digital agriculture platform that provides financial support to the farmers by allowing those outside the agriculture industry to sponsor individual farms.

Several other agritech start-ups across the continent, such as Ghana-based Farmerline and AgroCenta, and Nigeria-based Kitovu have also launched data-driven mobile application for farmers. These technology solutions are proving to be a boon for agriculture sector in Africa, helping improve the overall efficiency and productivity.

Agritech development is concentrated in Kenya and Nigeria

But, when it comes to first adopting the newest technologies and starting an agritech business in agriculture, Kenya and Nigeria have been leading in the adoption of new agritech solutions, accounting for a significant share of agritech start-up across Africa. Kenya has played a pioneering role in bringing agritech in Africa since 2010-2011, when the first wave of agritech start-ups began to bring new niche innovations. Currently, Kenya accounts for 25% of all the agritech start-ups in Africa, and the development is progressing rapidly, thanks to the country’s advancement in technology, high smartphone penetration, and relatively widespread internet access.

Similarly, Nigeria too has sailed the boat of success in agritech start-ups since 2015, and now it accounts for 23.2% of total agritech start-ups in Africa, with include major players such as Twiga Foods, Apollo Agriculture, Agrikore, and Tulaa. The growing inclination amongst Nigerian farmers towards using digital tools in agriculture sector has further pushed the rapid development in agritech sector in the country.

Other countries have also shown potential for agritech development, though it is still in the initial stages of becoming mainstream in their agriculture sectors. Ghana has encouraged several start-ups to launch different technology innovations for making agriculture more sustainable, while South Africa, Uganda, and Zimbabwe have also witnessed the rise in agritech start-ups over the years with newer technologies for agriculture sector.

Recent investments highlight the agritech potential

The agriculture technologies in Africa got the boost from the increased private funding. According to a report by Disrupt-Africa released in 2018, there has been a total investment of US$19 million in agritech sector since 2016. These investments have largely focused on funding agritech start-ups working on bringing innovative agriculture technologies. Also, according to the same report, the number of agritech start-ups rose by 110% from 2016 to 2018.

Some of the recent investments in the agritech sector include Kenya’s Twiga Foods, a B2B food distribution company, which raised US$30 million from investors led by Goldman Sachs in October 2019. The company aims to set-up a distribution centre in Nairobi to offer better supply chain services, while also expanding to more cities in Kenya, including Mombasa.

In December 2019, Kenya-based agritech start-up Farmshine, also raised US$25 million in funding from US-based Gray Matter’s Capital coLabs (GMC coLabs), to expand its operations in Malawi. GMC coLabs also invested US$1 million in another Kenyan B2B agritech start-up Taimba in July 2019. Taimba provides a mobile-based cashless platform connecting smallholder farmers to urban retailers. The investment was focused on strengthening Taimba’s infrastructure and increase the delivery logistics to cater to new markets.

Cellulant, a leading pan-African digital payments service provider that offers a real-time payment platform to farmers, also raised US$47.5 million from a consortium of investors in May 2018, which is the largest investment in the African tech industry till date. Cellulant also plans to channel a significant portion of funds into its Agrikore subsidiary, an agritech start-up dealing with blockchain based smart-contracting, payments, and marketplace system.

EOS Perspective

African agritech is expected to witness high growth in future. According to a CTA report on Digitalization for Agriculture (D4Ag) published in 2018, digital agriculture solutions are likely to reach 60-100 million smallholder famers, while generating annual revenues of nearly US$320- US$470 million by the end of 2020.

Adoption and use of innovative technologies such as remote sensing, diagnostics, IoT sensors for digitalization of agriculture is steadily moving from experimental stage to full-scale deployment, contributing to the data revolution in agriculture, while also unlocking new business models and opportunities.

Apart from these, blockchain is gaining prominence, and finding applications in the agriculture sector in Africa. This technology has the potential to significantly impact the agriculture sector, which we will discuss in the second part of our series on Agritech in Africa.

However, lack of affordability and knowledge to access such technologies, especially by small-scale farmers, has restricted the growth and reachability of these solutions. With the need to educate farmers and make such technology affordable and viable, it is likely that it may take at least 5-7 years before these technologies become truly mainstream in the continent.

A disparity of investments has been observed among the countries in the region. Over the years, countries such as Kenya, Nigeria, and Ghana have experienced a strong growth in terms of private investments, while other countries are left wanting. Investors have prioritized easy-to-reach markets in Africa, leaving behind the lower-income markets, resulting in agritech becoming less sustainable and scalable in these markets. However, several other African countries have shown the appetite to adopt agritech solutions, and offer significant potential.

This requires an intervention and participation from both governments and private investors, which can help improve scalability of agriculture technologies in the region. Implementation of farming digital literacy, public-private partnerships, and increased private sector investments in agritech enterprises can help the agritech industry experience a consistent and higher success rate, thus bringing the agriculture technology to a mainstream at faster pace.

24/07/2019by EOS Intelligence EOS Intelligence No Comments

Infographic: Google’s Tech Initiatives Transforming Industries

2.5kviews

Google, beyond being the leading search engine worldwide, is also one of the largest and most innovative companies. Through its innovations, Google along with other Alphabet companies (parent company of Google and its subsidiaries) is transforming various industries by empowering them with technology. Its solutions have reached diverse industries such as agriculture, manufacturing, healthcare, energy, and fishing, among others.

Innovation has always been at the core of Google’s strategy and it is bringing artificial intelligence (AI), machine learning, augmented reality, robotics, among others to shape various industries. It has introduced surgical robots to medicine, Google glass to manufacturing, AI-enabled programs to energy, among various other solutions that are revolutionizing these industries. We are taking a look at where Google has already left its innovative footprint.

Alphabet companies included in the infographic:

Verily – Alphabet’s key research organization dedicated to the study of life sciences
Verb Surgical – A joint venture between Johnson & Johnson and Verily
DeepMind – Alphabet’s artificial intelligence company
Global Fishing Watch – An organization founded by Google in partnership with Oceana and SkyTruth

06/03/2019by EOS Intelligence EOS Intelligence No Comments

Slowly but Surely – Insurance Realizes AI’s Value

1.2kviews

Several sectors, such as banking, F&B, automotive, and healthcare have seen major transformations at the hands of artificial intelligence (AI) ‒ we discussed benefits of AI in fast food industry in our previous article – Artificial Intelligence Finds its Way into Your Favorite Fast Food Chain in November 2017. AI has become an integral part of a large number of industries, providing new solutions and facilitating greater back-end efficiency as well as customer engagement and management. Insurance sector, on the other hand, has been largely slow to react to this disruptive trend. In 2017, only about 1.3% of insurance companies invested in AI (as compared with 32% insurance companies that invested in software and information technologies). However, this is expected to change as insurance companies have begun to realize the untapped potential that AI unearths in all aspects of their business, i.e. policy pricing, customer purchase experience, application processing and underwriting, and claim settlement.

Insurance industry has been one of the sectors that have operated in their traditional form for several decades, without undergoing much of substantial transformation. This is also one of the reasons why the insurance sector has been relatively late in jumping on the AI bandwagon.

Artificial intelligence, which has significantly transformed the way several industries such as automotive, healthcare, and manufacturing operate, also presents a host of benefits to the insurance sector. Moreover, it is expected to drive savings not only for insurance companies but also brokers and policy holders.

Streamlining internal processes

AI has the ability to streamline several internal processes within insurance companies. There is a host of duplicating business operations in the insurance sector. Automation and digitization can result in about 40% cost cutting, and this can be achieved by automating about 30% of the operations.

This can be seen in the case of Fukoku Mutual Life Insurance. In 2017, this Tokyo-based insurance company replaced 34 employees with IBM’s Watson Explorer AI system that can calculate payouts to policyholders in faster and more precise manner. The company expects to boost productivity by 30% and is expected to save close to US$1.26 million (JPY 140 million) in the first year of operations. To put this in a perspective, the AI system cost the company, US$1.8 million (JPY 200 million), and its maintenance is expected to cost US$130,000 (JPY 15 million) per year. Therefore, Fukoku seems optimistic about achieving its return on investment within less than two years of installing the AI system.

In addition to providing automation of processes, AI can bring out disruptive transformation throughout the insurance value chain. Some of the most substantial benefits of using AI in the insurance sector are expected to be seen in policy pricing, offering of personalized insurance plans, as well as claim management.

Policy pricing

Traditionally, insurance companies used to price their policies by creating risk pools based on statistical sampling, thereby all insurance policies were based on proxy data.

AI is transforming this by moving policy pricing analysis from proxy data to real-time source data. Internet of Things (IoT) device sensors, such as telematics and wearable sensor data, enable insurance firms to price coverage based on real events and real-time data of the individuals that they are insuring.

An example of this is usage-based or pay-per-mile auto insurance, wherein a telematics sensor box (a black box for a car), is installed into a car to track information such as speed, driving distances, breaking habits, and other qualitative and quantitative driving data. Using this data, insurance companies can offer a customized policy to the car owner, charging lower premium from safe drivers or offering less-used cars the pay-per-mile option. It also helps insurance companies charge suitable premium from reckless drivers and long-distance drivers.

In February 2017, UK-based mobile network brand, O2, expanded into the auto insurance space with a telematics product called the O2 Drive. The device tracks different aspects of a user’s driving habits and offers discounts and personalized insurance policies based on it. The company is positioning its products to attract teen and young drivers as they are most likely to be open to sharing their driving data.

In addition to auto insurance, IoT devices such as wearable devices and smart home solutions also help in setting policy pricing in health and home insurance. US-based Beam Insurance Services uses a smart toothbrush to offer dental insurance. The company uses data accrued from the smart toothbrush, such as number of times a person brushes their teeth, duration of brushing, etc., to offer a personalized insurance policy. It claims to offer up to 25% lower rates in comparison with its competitors.

In another example, UK-based Neos Ventures offers IoT-powered home insurance based on a smart home monitoring and emergency assistance device. The device and its accompanying app helps users reduce instances of fire and water-based damages as well as break-ins and thefts. The premise of the company is that if they can successfully reduce the chances of any mishaps, they can offer cheaper premiums to the insured.

While IoT devices can greatly personalize insurance pricing, the largest caveat to the success of this pricing mechanism remains that customers must be willing to share their personal data with insurance providers to attain savings in the form of lower premium. As per Deloitte – EMEA Insurance Data Analytics Study 2017, about 40% of customers surveyed seemed open to track their behavior and share the data with insurers for more accurate premiums for health insurance, while 38% and 48% customers were open to tracking and sharing data in case of home and auto insurance, respectively.

Customer purchase experience and underwriting of applications

The relationship between an insurance agent and the customer is an extremely important one for insurance companies. Many times the customer is dissatisfied with its interaction/experience with the insurance agent as they feel that the agent does not have their best interest at heart or the agent is not available for them as and when required.

This issue is effectively addressed with the use of AI-powered chatbots or virtual assistants. Advanced chatbots use image recognition and social data to personalize sales conversations and provide a better customer experience. Thus, agents and insurance representatives are being replaced by chatbots, which deliver faster and more efficient customer experience.

ZhongAn, a China-based pure online insurance company uses chatbots for 97% of its customer queries without any human involvement. It also uses AI to offer innovative insurance products, such as cracked mobile screen insurance. It uses image recognition technology to detect whether the image shows the mobile screen is cracked or intact. It can also decipher if the picture has been photoshopped or altered to ensure the claim is genuine. Since its inception in 2013, the company has sold about 8 billion policies to 500 million customers (these include cracked mobile insurance as well as the company’s other popular products).

To blend the human experience with chatbots, companies have started branding their chatbots with human names. New York-based P2P insurance company, Lemonade, uses exclusively chatbots named Maya and Jim to interact with customers and create personalized insurance options in less than a minute within the Lemonade app. The chatbots Maya and Jim are alter-egos of the company’s real-life employees with the same names.

Similarly, in December 2016, ICICI Lombard General Insurance launched a chatbot called MyRA. Within six months of operations the virtual assistance platform sold 750 policies without any human intervention, while it was used by 60,000 consumers for queries.

In addition to elevating customer’s purchase experience, AI also helps in reducing insurance underwriting/processing time and ensuring higher quality. The underwriting process traditionally has a range of manual tasks that make the process slow and also prone to human errors. However, AI helps achieve quicker and more reliable data analysis. AI tools such as Machine Learning and Natural Language Processing (NLP) help underwriters scan a customer’s social profile to gather important data, trends, and behavioral patterns that can result in more accurate assessment of the application.

New-York based Haven Life (a subsidiary of MassMutual), leverages AI technology to underwrite its life insurance policies. It requires its customers to submit a 30-question application (which is more conversational in nature as compared with the detailed traditional life insurance forms) and upload few documents such as medical records, motor vehicle driving records, etc. The AI technology analyzes the provided information along with historical life insurance data and asks additional questions if required. In several cases, it also offers coverage without the mandated medical test. Through AI, the company has reduced its underwriting time from the typical 1-2 weeks to as low as 20 minutes.

Claim management

AI can play a significant role in two of the most critical aspects of claim management, i.e. the time to settle a claim and fraud detection.

The time to settle a claim is one of the performance metrics that customers care most about. Using AI, companies can expedite the claim process. Chatbots are used to address the First Notice of Loss (FNOL), wherein customers submit their claims by sending pictures of the damaged goods along with answering few questions. The chatbot then processes the claim and assesses the extent of loss and its authenticity, to determine the correct amount for claim settlement.

Lemonade set a world record in December 2016 by settling a claim using its AI bot, Jim, in only three seconds. The AI bot reviewed the claim, cross-referenced it against the policy, ran several anti-fraud algorithms, approved the claim, sent wiring instructions to the bank, and informed the customer in the three-second window.

Another interesting area of application is in agriculture, where machine learning can also help quickly analyze claims (pertaining to loss spread over a wide area) using satellite imaging, which would otherwise take humans significantly greater time and costs to ascertain.

As mentioned earlier, AI can bring massive savings to insurance firms by reducing fraudulent claims. As per US-based Coalition Against Insurance Fraud (CAIF) estimates, insurance carriers lose about US$80 billion annually in fraudulent claims. AI technologies provide insurance firms with real-time data to identify duplicate and inflated claims as well as fake diagnoses.

In addition, many companies use AI to run algorithms on historical data to identify sequences and patterns of fraudulent claims to identify traits and trends that may be missed by the human eye during the initial stages of claim processing.

According to CAIF, in November 2016, about 75% of insurance firms used automated fraud detection systems to detect false claims. Paris-based Shift Technologies is one of the leading players in this domain, claiming to have a 250% better fraud identification rate as compared with the market average. The company had analyzed more than 100 million claims from its inception in 2013 up till October 2017.

EOS Perspective

There is no denying that AI has the capability to transform the insurance industry (as it has transformed many other industries). Although, initially slow at reacting to the AI trend, insurance companies have realized its potential.

As per an April 2017 Accenture survey, about 79% of the insurance executives believed that AI will revolutionize the way insurers gain information from and interact with their customers. This is also visible in the recent level of investments made in AI by the insurance sector. TCS’s Global Trend Study on AI 2017 stated that the insurance sector outspent all the other 12 sectors surveyed (including travel, consumer packaged goods, hospitality, media, etc.) by investing an average of US$124 million annually in AI systems. The cross industry average of the 13 sectors stood at US$70 million.

Thus, it is very important for insurance players to get on board the AI trend now. Since they are already late (in comparison to some other industries) in reacting to the trend, it is critical that they adapt to it to remain relevant and competitive.

However, the key barrier to AI implementation are the complex and outdated legacy systems that hold back innovation and digitization. The companies that do not manage to implement tech innovations in their legacy systems due to high cost might just be acting penny wise, pound foolish.

Who We Are

Connect With Us

Work With EOS

Who We Are

Connect With Us

Work With EOS

How could ChatGPT help businesses?

ChatGPT’s output may not be always accurate

ChatGPT is not the first or only AI chatbot

ChatGPT-styled chatbots to power search engines

EOS Perspective

AI-based virtual simulation models used for effective R&D processes

General Motors, Volkswagen, and Hyundai use AI for smart manufacturing

Companies provide more ADAS features amidst increasing competition

Volkswagen and Porsche use AI in automotive marketing and sales

Automakers integrate AI-assisted voice assistants into cars

Integration and technological challenges hamper the adoption of AI

EOS Perspective

Increasing capital investment signals rising interest in AI in healthcare diagnostics

AI advantages help answer the needs in healthcare diagnostics

Black-box AI: A source of challenges to AI implementation in healthcare diagnostics

Financial challenges

Technological challenges

Mental and psychological challenges

EOS Perspective

Personalization through AI

Virtual makeup through AR and VR

Smart devices

EOS Perspective

State of blockchain implementation in agriculture in Africa

Blockchain to promote transparency across agriculture sector

Blockchain-based platforms to improve farmer and buyer collaboration

Smart contracts to transform agriculture finance and insurance

Improving mobile internet access to boost blockchain implementation

EOS Perspective

Start-ups lead agritech development in Africa

Agritech development is concentrated in Kenya and Nigeria

Recent investments highlight the agritech potential

EOS Perspective

Alphabet companies included in the infographic:

Streamlining internal processes

Policy pricing

Customer purchase experience and underwriting of applications

Claim management

EOS Perspective